evidence for single top quark production at dØ...3/13/2007 shabnam jabeen (bu) 6 dataset dØ has...
TRANSCRIPT
Evidence forEvidence forSingle Top Quark Production Single Top Quark Production at DØat DØ
ShabnamShabnam JabeenJabeenBoston University Boston University
HEP Seminar, KUHEP Seminar, KUMarch 13, 2007March 13, 2007
3/13/2007 Shabnam Jabeen (BU) 2
Top QuarksTop QuarksSpin 1/2 Spin 1/2 fermionfermion, charge +2/3, charge +2/3WeakWeak--isospinisospin partner of the bottom partner of the bottom quarkquark~40x heavier than its partner~40x heavier than its partnerHeaviest known fundamental particleHeaviest known fundamental particle
ttop
Produced mostly in Produced mostly in tttt pairs at the pairs at the TevatronTevatron85% 85% tttt,, 15% 15% gggg
200
150
100
50
0
up
dow
n
stra
nge
char
m
botto
m top
[GeV
]
MtopMtop = 171.4 = 171.4 ±± 2.1 2.1 GeVGeV
Cross section = 6.8 Cross section = 6.8 ±± 0.6 0.6 pbpb at NNLOat NNLODiscovered by D0 and CDF in 1995Discovered by D0 and CDF in 1995
3/13/2007 Shabnam Jabeen (BU) 3
The DØ ExperimentThe DØ Experiment
FermilabTevatron
Top quarks observed by Top quarks observed by DØ and CDF in 1995 with DØ and CDF in 1995 with ~50 ~50 pbpb––11 of dataof data
D0
CDF
3/13/2007 Shabnam Jabeen (BU) 4
The DØ ExperimentThe DØ ExperimentTop quarks observed by Top quarks observed by DØ and CDF in 1995 with DØ and CDF in 1995 with ~50 ~50 pbpb––11 of dataof data
3/13/2007 Shabnam Jabeen (BU) 5
The DØ ExperimentThe DØ ExperimentTop quarks observed by Top quarks observed by DØ and CDF in 1995 with DØ and CDF in 1995 with ~50 ~50 pbpb––11 of dataof data
3/13/2007 Shabnam Jabeen (BU) 6
DatasetDatasetDØ has more than 2 DØ has more than 2 fbfb––11 on tape, on tape, Many thanks to the Many thanks to the FermilabFermilab accelerator division!accelerator division!This analysis uses 0.9 This analysis uses 0.9 fbfb––11 of data collected from 2002 to of data collected from 2002 to 20052005
3/13/2007 Shabnam Jabeen (BU) 7
Single Top OverviewSingle Top Overview
s-channel: “tb”σσNLONLO = = 0.88 0.88 ±± 0.110.11 pbpb
t-channel: “tqb”σσNLONLO = = 1.98 1.98 ±± 0.250.25 pbpb
Experimental results Experimental results (95% C.L.)(95% C.L.)DØDØ tbtb < 5.0 < 5.0 pbpb (370 (370 pbpb––11))
CDFCDF tbtb < 3.2 < 3.2 pbpb (700 (700 pbpb––11))DØDØ tqbtqb < 4.4 < 4.4 pbpb (370 (370 pbpb––11))CDFCDF tqbtqb < 3.1 < 3.1 pbpb (700 (700 pbpb––11))
CDFCDF tb+tqbtb+tqb < 2.7 < 2.7 pbpb LikelihoodsLikelihoods (960 (960 pbpb––11))tb+tqbtb+tqb < 2.6 < 2.6 pbpb Neural networksNeural networkstb+tqbtb+tqb = 2.7 = 2.7 pbpb Matrix elementsMatrix elements (significance of 2.3 (significance of 2.3 σσ))–1.3
“tW production”σσNLONLO = = 0.210.21 pbpb
(Too small to see at the (Too small to see at the TevatronTevatron))
+1.5+1.5––1.31.3
3/13/2007 Shabnam Jabeen (BU) 8
Why Look for Single Top?Why Look for Single Top?Study Study WtbWtb coupling in top productioncoupling in top production
MeasureMeasure ||VVtbtb| | directly (more later)directly (more later)Test Test unitarityunitarity of CKM matrixof CKM matrixAnomalous Anomalous WtbWtb couplingscouplings
Cross sections sensitive to new physicsCross sections sensitive to new physicsss--channel: resonances (heavy channel: resonances (heavy WW′′ boson, charged Higgs boson, boson, charged Higgs boson, KaluzaKaluza--Klein excited Klein excited WWKKKK, , technipiontechnipion, etc.), etc.)tt--channel: flavorchannel: flavor--changing neutral currents (changing neutral currents (t t –– Z / Z / γγ / g / g –– c / uc / u couplings)couplings)Fourth generation of quarksFourth generation of quarks
Top propertiesTop propertiesPolarized top quarks Polarized top quarks –– spin correlations measurable in decay productsspin correlations measurable in decay productsMeasure top quark partial decay width and lifetimeMeasure top quark partial decay width and lifetimeCP violation (same rate for top and CP violation (same rate for top and antitopantitop?)?)
Similar (but easier) search than for Similar (but easier) search than for WHWH associated Higgs productionassociated Higgs productionBackgrounds the same Backgrounds the same –– must be able to model them successfullymust be able to model them successfullyTest of techniques to extract a small signal from a large backgrTest of techniques to extract a small signal from a large backgroundound
3/13/2007 Shabnam Jabeen (BU) 9
Event SelectionEvent Selection
One isolated electron or One isolated electron or muonmuonElectron Electron ppTT > 15 > 15 GeVGeV, , ||ηη| < 1.1| < 1.1MuonMuon ppTT > 18 > 18 GeVGeV, , ||ηη| < 2.0| < 2.0
Missing transverse energyMissing transverse energyEETT > 15 > 15 GeVGeV
One bOne b--tagged jet and atagged jet and at least one more jett least one more jet22––4 jets with 4 jets with ppTT > 15 > 15 GeVGeV, |, |ηη| < 3.4| < 3.4Leading jet Leading jet ppTT > 25 > 25 GeVGeV, |, |ηη| < 2.5| < 2.5Second leading jet Second leading jet ppTT > 20 > 20 GeVGeV
3/13/2007 Shabnam Jabeen (BU) 10
Signal and Background ModelsSignal and Background ModelsSingle top quark signals modeled using SINGLETOPSingle top quark signals modeled using SINGLETOP
By Moscow State University theorists, based on COMPHEPBy Moscow State University theorists, based on COMPHEPReproduces NLO Reproduces NLO kinematickinematic distributionsdistributionsPYTHIA for PYTHIA for partonparton hadronizationhadronization
tttt pair backgrounds modeled using ALPGENpair backgrounds modeled using ALPGENPYTHIA for PYTHIA for partonparton hadronizationhadronizationPartonParton--jet matching algorithm used to avoid doublejet matching algorithm used to avoid double--counting final counting final statesstatesNormalized to NNLO cross sectionNormalized to NNLO cross section18% uncertainty includes component for top mass18% uncertainty includes component for top mass
MultijetMultijet background modeled using data with a nonbackground modeled using data with a non--isolated lepton and jetsisolated lepton and jetsNormalized to data before Normalized to data before bb--tagging (together with tagging (together with WW+jets +jets background)background)
WW+jets Background+jets BackgroundWW+jets background modeled using ALPGEN+jets background modeled using ALPGENPYTHIA for PYTHIA for partonparton hadronizationhadronizationPartonParton--jet matching algorithm used to avoid doublejet matching algorithm used to avoid double--counting final statescounting final statesWbbWbb and and WccWcc fractions from data to better represent higherfractions from data to better represent higher--order effectsorder effects30% uncertainty for differences in event kinematics and assuming30% uncertainty for differences in event kinematics and assuming equal for equal for WbbWbband and WccWccWW+jets normalized to data before +jets normalized to data before bb--tagging (with tagging (with multijetmultijet background)background)ZZ+jets, +jets, dibosondiboson backgrounds very small, included in backgrounds very small, included in WW+jets via normalization+jets via normalization
11
Event Yields Before Event Yields Before bb--TaggingTaggingW Transverse MassW Transverse Mass
2 jets
ElectronsElectrons MuonsMuons
3 jets
4 jets
Signal acceptances: Signal acceptances: tbtb = 5.1%, = 5.1%, tqbtqb = 4.5%= 4.5%
S:B ratio for S:B ratio for tb+tqbtb+tqb = 1:180= 1:180
Need to improve S:B to have a hope of Need to improve S:B to have a hope of seeing a signal seeing a signal →→ select only events with select only events with bb--jets in themjets in them
12
bb--Jet IdentificationJet Identification
Ann Heinson (UC Riverside)
13
Separate b-jets from light-quark and gluon jets to reject most W+jets background
DØ uses a neural network algorithm7 input variables based on impact parameter and reconstructed vertex
Operating point:b-jet efficiency ≈ 50%c-jet efficiency ≈ 10%light-jet effic. ≈ 0.5%
3/13/2007 Shabnam Jabeen (BU) 14
Event Yields after Event Yields after bb--TaggingTagging
Signal acceptances: tb = (3.2 ± 0.4)%, tqb = (2.1 ± 0.3)%
Signal:background ratios for tb+tqb are 1:10 to 1:50Most sensitive channels have 2jets/1tag, S:B = 1:20
Single top signal is smaller than total background uncertaintycounting events is not a sensitive enough methoduse a multivariate discriminant to separate signal from background
3/13/2007 Shabnam Jabeen (BU) 15
Search Strategy SummarySearch Strategy Summary
Maximize the signal acceptanceMaximize the signal acceptanceParticle ID definitions set as loose Particle ID definitions set as loose as possible (i.e., highest as possible (i.e., highest efficiency, separate signal from efficiency, separate signal from backgrounds with fake leptons backgrounds with fake leptons later)later)Transverse momentumTransverse momentumthresholds set low, thresholds set low, pseudorapiditiespseudorapidities widewideAs many decay channels used as As many decay channels used as possible possible –– this analysis shown in this analysis shown in red boxred boxChannels analyzed separately Channels analyzed separately since S:B and background since S:B and background compositions differcompositions differ
Separate signal from background Separate signal from background using multivariate techniquesusing multivariate techniques
3/13/2007 Shabnam Jabeen (BU) 16
12 Analysis Channels 12 Analysis Channels W Transverse Mass
2 jets
Electrons
3 jets
4 jets
1 tag 2 tagsMuons
1 tag 2 tags
3/13/2007 Shabnam Jabeen (BU) 17
Systematic UncertaintiesSystematic UncertaintiesUncertainties are assigned for each signal and background componUncertainties are assigned for each signal and background component ent in all analysis channelsin all analysis channels
Most systematic uncertainties apply only to normalizationMost systematic uncertainties apply only to normalization
Two sources of uncertainty also affect the shapes of distributioTwo sources of uncertainty also affect the shapes of distributionsnsjet energy scalejet energy scaletagtag--rate functions for rate functions for bb--tagging MC eventstagging MC events
Correlations between channels and sources are taken into accountCorrelations between channels and sources are taken into account
Cross section uncertainties are Cross section uncertainties are dominated by the statistical uncertaintydominated by the statistical uncertainty,, the the systematic contributions are all smallsystematic contributions are all small
Source of UncertaintySource of Uncertainty SizeSizeTop pairs normalizationTop pairs normalization 18%18%W+jets & W+jets & multijetsmultijetsnormalizationnormalization
1818––28%28%
Integrated luminosityIntegrated luminosity 6%6%Trigger modelingTrigger modeling 33––6%6%Lepton ID correctionsLepton ID corrections 22––7%7%Jet modelingJet modeling 22––7%7%Other small componentsOther small components Few %Few %Jet energy scaleJet energy scale 11––20%20%Tag rate functionsTag rate functions 22––16%16%
3/13/2007 Shabnam Jabeen (BU) 18
Measuring a Cross SectionMeasuring a Cross Section
NbkgdsNbkgds = 6 (= 6 (ttllttll, , ttljttlj, , WbbWbb, , WccWcc, , WjjWjj, , multijetsmultijets), ), NbinsNbins = 12 = 12 chanschans x 100 bins = 1,200x 100 bins = 1,200Cross section obtained from peak position of Bayesian posterior Cross section obtained from peak position of Bayesian posterior probability densityprobability densityShape and normalization systematic uncertainties treated as nuisShape and normalization systematic uncertainties treated as nuisance parametersance parametersCorrelations between uncertainties are properly accounted forCorrelations between uncertainties are properly accounted forSignal cross section prior is nonSignal cross section prior is non--negative and flatnegative and flat
18
3/13/2007 Shabnam Jabeen (BU) 19
Final Analysis StepsFinal Analysis Steps
We have selected 12 independent sets of data for final analysisWe have selected 12 independent sets of data for final analysis
Background model gives good representation of data in ~90 Background model gives good representation of data in ~90 variables in every channelvariables in every channel
Calculate Calculate discriminantsdiscriminants that separate signal from backgroundthat separate signal from backgroundBoosted decision treesBoosted decision treesMatrix elementsMatrix elementsBayesian neural networksBayesian neural networks
Check Check discriminantdiscriminant performance using data control samplesperformance using data control samples
Use ensembles of pseudoUse ensembles of pseudo--data to test validity of methodsdata to test validity of methods
Calculate cross sectionsCalculate cross sections using binned likelihood fits ofusing binned likelihood fits of(floating) signal + (fixed) background to data(floating) signal + (fixed) background to data
3/13/2007 Shabnam Jabeen (BU) 20
Testing with PseudoTesting with Pseudo--DataDataTo verify that the calculation methods work as expected, we testTo verify that the calculation methods work as expected, we test them using them using several sets (“ensembles”) of pseudoseveral sets (“ensembles”) of pseudo--datadataWonderful tool to test the analyses! Like running DØ many 1,000Wonderful tool to test the analyses! Like running DØ many 1,000’s of times’s of timesSelect subsets of events from total pool of MC eventsSelect subsets of events from total pool of MC events
Randomly sample a Poisson distribution to simulate statistical fRandomly sample a Poisson distribution to simulate statistical fluctuationsluctuationsBackground yields fluctuated according to uncertainties to reproBackground yields fluctuated according to uncertainties to reproduce duce correlations between components from normalizationcorrelations between components from normalization
Ensembles we used:Ensembles we used:ZeroZero--signal ensemble, signal ensemble, σσ((tb+tqbtb+tqb) = 0 ) = 0 pbpbSM ensemble, SM ensemble, σσ((tb+tqbtb+tqb) = 2.9 ) = 2.9 pbpb““MysteryMystery”” ensembles, ensembles, σσ((tb+tqbtb+tqb) = ? ) = ? pbpbMeasured Measured XsecXsec ensemble, ensemble, σσ((tb+tqbtb+tqb) = ) = σσmeasmeas
Each pseudoEach pseudo--dataset is like one Ddataset is like one DØØ experiment with 0.9 experiment with 0.9 fbfb––1 of 1 of ““datadata””,,up to 68,000 pseudoup to 68,000 pseudo--datasets per ensembledatasets per ensemble
20
3/13/2007 Shabnam Jabeen (BU) 21
21
SignalSignal--Background SeparationBackground Separationusing Decision Treesusing Decision Trees
MachineMachine--learning technique, widely used in social sciences, learning technique, widely used in social sciences, some use in HEPsome use in HEPIdea: recover events that fail criteria in cutIdea: recover events that fail criteria in cut--based analysesbased analysesStart at first “node ” with “training sample” of 1/3 of all Start at first “node ” with “training sample” of 1/3 of all signal and background eventssignal and background events
For each variable, find splitting value with best separation For each variable, find splitting value with best separation between two children (mostly signal in one, mostly background between two children (mostly signal in one, mostly background in the other)in the other)Select variable and splitting value with best separation to Select variable and splitting value with best separation to produce two “branches” with corresponding events, (F)ailed and produce two “branches” with corresponding events, (F)ailed and (P)assed (P)assed cutcutRepeatRepeat recursively on each noderecursively on each node
Stop when improvement stops or when too few events are left Stop when improvement stops or when too few events are left (100)(100)Terminal node is called a “leaf ” withTerminal node is called a “leaf ” with
purity = purity = Nsignal/(Nsignal+NbackgroundNsignal/(Nsignal+Nbackground))
Run remaining 2/3 events and data through tree to derive resultsRun remaining 2/3 events and data through tree to derive resultsDecision tree output for each event = leaf purity Decision tree output for each event = leaf purity (closer to 0 for background, nearer 1 for signal)(closer to 0 for background, nearer 1 for signal)
3/13/2007 Shabnam Jabeen (BU) 22
Boosting the Decision TreesBoosting the Decision TreesBoosting is a recently developed technique that improves any weak classifier (decision tree, neural network, etc)Boosting averages the results of many trees, dilutes the discrete nature of the output, improves the performanceThis analysis:
Uses the “adaptive boosting algorithm”:Train a tree TkCheck which events are misclassified by TkDerive tree weight wkIncrease weight of misclassified eventsTrain again to build Tk+1Boosted result of event i :
20 boosting cyclesTrained 36 sets of trees: (tb+tqb, tb, tqb) x (e,μ) x (2,3,4 jets) x (1,2 b-tags)Separate analyses for tb and tqb allow access to different types of new physicsSearch for tb+tqb has best sensitivity to see a signal
T (i) = wkTk (i)n=1Ntree∑
Before boosting After boosting
3/13/2007 Shabnam Jabeen (BU) 23
Decision Tree VariablesDecision Tree Variables
49 input variables - Same list of variables used for all analysis channelsAdding more variables does not degrade the performanceReducing the number of variables always reduces sensitivity of the analysis
3/13/2007 Shabnam Jabeen (BU) 24
Decision Tree Cross ChecksDecision Tree Cross ChecksSelect two background-dominated samples:
“W+jets”: = 2 jets, HT(lepton, ET, alljets) < 175 GeV, =1 tag“tt”: = 4 jets, HT (lepton, ET, alljets) > 300 GeV , =1 tag
Observe good data-background agreement“W+jets”
Electrons
“tt”
Muons
24
3/13/2007 Shabnam Jabeen (BU) 25
Decision Tree VerificationDecision Tree VerificationUse “mystery” ensembles with many different signal assumptions
Measure signal cross section using decision tree outputs
Compare measured cross sections to input ones
Observe linear relation close to unit slope
Input xsec
3/13/2007 Shabnam Jabeen (BU) 26
SignalSignal--Background SeparationBackground Separationusing Matrix Elementsusing Matrix Elements
Use the 4-vectors of all reconstructed leptons and jetsUse matrix elements of main signal and background Feynman diagrams to compute an event probability density for signal and background hypothesesGoal: calculate a discriminant:
Define PSignal as a normalized differential cross section:
Performed in 2-jets and 3-jets channels onlyNo matrix element for tt so no discrimination between signal and top pairs yetMatrix element verification with ensembles shows good linearity, unit slope, near-zero intercept
3/13/2007 Shabnam Jabeen (BU) 27
Matrix Element Cross ChecksMatrix Element Cross ChecksSelect two background-dominated samples:
“Soft W+jets”: = 2 jets, HT(lepton, ET, alljets) < 175 GeV, =1 tag“Hard W+jets”: = 2 jets, HT(lepton, ET, alljets) > 300 GeV, =1 tag
Observe good data-background agreement
“Soft W+jets”
27
“Hard W+jets”
Full range High end Full range High end
tb
tq
3/13/2007 Shabnam Jabeen (BU) 28
SignalSignal--Background SeparationBackground Separationusing Bayesian Neural Networksusing Bayesian Neural Networks
Bayesian neural networks improve on this technique:Average over many networks weighted by the probability of each network given the training samplesLess prone to over-trainingNetwork structure is less important – can use larger numbers of variables and hidden nodes
For this analysis:24 input variables (subset of 49 used by decision trees)40 hidden nodes, 800 training iterationsEach iteration is the average of 20 training cyclesOne network for each signal (tb+tqb, tb, tqb) in each of the 12 analysis channels
Bayesian neural network verification with ensembles shows good linearity, unit slope, near-zero intercept
Neural networks use many input variables, train on signal and background samples, produce one output discriminant
Network output
Network output
tqb
Wbb
3/13/2007 Shabnam Jabeen (BU) 29
Statistical AnalysisStatistical AnalysisBefore looking at the data, we want to know two things:
By how much can we expect to rule out a background-only hypothesis?Find what fraction of the ensemble of zero-signal pseudo-datasets give a cross section at least as large as the SM value, the “expected p-value”For a Gaussian distribution, convert p-value to give “expected signficance”
What precision should we expect for a measurement?Set value for “data” = SM signal + background in each discriminant bin (non-integer) and measure central value and uncertainty on the “expected cross section”
With the data, we want to know:How well do we rule out the background-only hypothesis?
Use the ensemble of zero-signal pseudo-datasets and find what fraction give a cross section at least as large as the measured value, the “measured p-value”Convert p-value to give “measured signficance”
What cross section do we measure?Use (integer) number of data events in each bin to obtain “measured cross section”
How consistent is the measured cross section with the SM value?Find what fraction of the ensemble of SM-signal pseudo-datasets give a cross section at least as large as the measured value to get “consistency with SM”
3/13/2007 Shabnam Jabeen (BU) 30
Expected ResultsExpected ResultsDecision Trees
1.9 %2.1 σ
2.7 pb
Expected p-valueExpected significanceExpected cross section +1.6
–1.4
Matrix Elements3.7 %1.8 σ
3.0 pb+1.8–1.5
Bayesian NNs9.7 %1.3 σ
3.2 pb+2.0–1.8
Decision Trees
Probability to rule outbackground-only
hypothesis
Zero-signal ensemble
SM =2.9 pb
MatrixElements
Zero-signalensemble
SM =2.9 pb
BayesianNeural
Networks
Zero-signalensemble
SM =2.9 pb
BayesianNeural Networks
Expectedresult
Matrix Elements
Expectedresult
Decision Trees
“Data” =SM signal +background
Expectedresult
3/13/2007 Shabnam Jabeen (BU) 31
Bayesian NN ResultsBayesian NN Results
σ(tb+tqb) = 5.0 ± 1.9 pbMeasured p-value = 0.89 %Measured significance = 2.4 σCompatibility with SM = 18%
BayesianNeural
Networks
Measuredresult
5.0 pb
BayesianNeural
Networks
Zero-signalensemble
Probabilityto rule outbackground-onlyhypothesis
BayesianNeural
Networks
SM-signalensemble
CompatibilityWith SM
5.0 pb
3/13/2007 Shabnam Jabeen (BU) 32
Matrix Element ResultsMatrix Element Results
σ(tb+tqb) = 4.6 pb
Measured p-value = 0.21 %Measured significance = 2.9 σCompatibility with SM = 21%
+1.8–1.5
MatrixElements
Measuredresult
4.6 pb
MatrixElements
Zero-signal
ensemble
Probabilityto rule outbackground-onlyhypothesis
MatrixElements
SM-signalensemble
CompatibilityWith SM
4.6 pb
3/13/2007 Shabnam Jabeen (BU) 33
Matrix Element ResultsMatrix Element Results
Discriminant output without and with signal component(all channels combined to “visualize” excess)
3/13/2007 Shabnam Jabeen (BU) 34
Decision Tree ResultsDecision Tree Results
σ(tb+tqb) = 4.9 ± 1.4 pb
Measured p-value = 0.035 %Measured significance = 3.4 σCompatibility with SM = 11%
DecisionTrees
Measuredresult
4.9 pb
DecisionTrees
Zero-signal
ensemble
Probabilityto rule outbackground-onlyhypothesis
DecisionTrees
SM-signalensemble
CompatibilityWith SM
4.9 pb
3/13/2007 Shabnam Jabeen (BU) 35
Decision Tree ResultsDecision Tree Results
Discriminant output (all channels combined) over the full range and a close-up on the high end
3/13/2007 Shabnam Jabeen (BU) 36
ME Event CharacteristicsME Event Characteristics
Mass (lepton,ET,btagged-jet) [GeV]
Q(lepton) x η(untagged-jet)
Mass (lepton,ET,btagged-jet) [GeV]
ME Discriminant < 0.4 ME Discriminant > 0.7
Q(lepton) x η(untagged-jet)
3/13/2007 Shabnam Jabeen (BU) 37
DT Event CharacteristicsDT Event Characteristics
Mass (lepton,ET,btagged-jet) [GeV]
W Transverse Mass [GeV]
DT Discriminant < 0.3
Mass (lepton,ET,btagged-jet) [GeV]
DT Discriminant > 0.65
W Transverse Mass [GeV]
3/13/2007 Shabnam Jabeen (BU) 38
Correlation Between MethodsCorrelation Between MethodsChoose the 50 highest events Choose the 50 highest events in each in each discriminantdiscriminant and and count overlapping eventscount overlapping events
Measure cross section in 400 pseudo-datasets of SM-signal ensemble and calculatecalculate linear correlation between each pair of results
Correlation between measured cross sectionsDT ME BNN
DT 100 % 39 % 57 %ME 100 % 29 %
BNN 100 %
Results from the three methods are consistent with each other
100 %BNN52 %100 %ME48 %58 %100 %DT
Overlap of signal-like events
100 %BNN46 %100 %ME56 %52 %100 %DTBNNMEDT
Ele
ctro
nsM
uons
3/13/2007 Shabnam Jabeen (BU) 39
CKM Matrix Element CKM Matrix Element VtbVtb
Weak interaction eigenstates and mass eigenstates are not the same: there is mixing between quarks, described by CKM matrix
In the SM, top must decay to W and d, s, or b quark
Constraints on Vtd and Vts give
If there is new physics, then
No constraint on VtbInteractions between top quark and gauge bosons are very interesting
3/13/2007 Shabnam Jabeen (BU) 40
Measuring |Measuring |VtbVtb||
AssumeSM top quark decay :Pure V–A : = 0CP conservation : = = 0
No need to assume only three quark families or CKM matrix unitarity(unlike for previous measurements using tt decays)
Measure the strength of the V–Acoupling, |Vtb |, which can be > 1
Additional theoretical uncertainties
tb tqbTop mass 13 % 8.5 %Scale 5.4 % 4.0 %PDF 4.3 % 10 %αs 1.4 % 0.01 %
Use the measurement of the single top cross section to make the first direct measurement of |Vtb|Calculate a posterior in |Vtb|2 (σ(tb, tqb) ∝ |Vtb|2)General form of Wtb vertex:
3/13/2007 Shabnam Jabeen (BU) 41
+0.6–0.5
First Direct Measurement of |First Direct Measurement of |VtbVtb||
|Vtbf1L| = 1.3 ± 0.2
0.68 < |Vtb| ≤ 1 at 95% C.L.(assuming f1L = 1)
|Vtbf1L|2
= 1.7+0.0–0.2
|Vtb|2 = 1.0
|Vtbf1L| = 1.3 ± 0.2
3/13/2007 Shabnam Jabeen (BU) 42
Challenging measurement – small signal hidden in huge complex backgroundMuch time spent on tool development (b-tagging) and background modelingThree multivariate techniques applied to separate signal from backgroundBoosted decision trees give result with 3.4 σsignificance
First direct measurement of |Vtb|
Result submitted to Physical Review LettersDoor is now open for studies of Wtb coupling and searches for new physics
Summary: Evidence forSummary: Evidence forSingle Top Quark Production at DØSingle Top Quark Production at DØ
Additional MaterialAdditional Material
3/13/2007 Shabnam Jabeen (BU) 44
Results for Results for tbtb and and tqbtqb SeparatelySeparately
DecisionTrees
Measuredresult for
s-channel tb
DecisionTrees
Measuredresult for
t-channel tqb
σ(tqb) = 4.2 pb
σ(tb) = 1.0 ± 0.9 pb
+1.8–1.4
3/13/2007 Shabnam Jabeen (BU) 45
Matrix Element MethodMatrix Element MethodFeynman DiagramsFeynman Diagrams
tb tq
2-jet channels
Wbb Wcg Wgg
3-jet channels
tbg tdb Wbbg
3/13/2007 Shabnam Jabeen (BU) 46
Matrix Element S:B SeparationMatrix Element S:B Separation46
tb discriminant tq discriminant2-jet channels